Periodic Report Summary - MOCCCASIN (Monitoring crops in continental climates through assimilation of satellite information)
Project context and objectives:
Information on the outlook of yield and production of crops over large regions is essential for government services dealing with import and export of food crops, for agencies with a role in food relief and for international organisations with a mandate in monitoring the world food production and trade. In 2007, unbalances in the global production of agricultural commodities due to co-occurrence of shortfalls in harvests caused the marked prices of agricultural commodities to peak. These events sparked widespread concern about global agricultural production. Moreover, increased demand for dairy, meat and biofuel products, which require cereals and oil crops, will increasingly compete with the demand for food crops. Given this background the need for a global monitoring system for agricultural production is undisputed and included in the priority themes for Group on Earth observations (GEO). Moreover, the importance was confirmed by the ministerial declaration of the G20 meeting of agricultural ministers in June 2011.
In Europe, the necessity for monitoring agricultural production and prediction of crop yield has long been recognised and implemented through the MARS crop yield forecasting system (MCYFS) operated by the Joint research centre (JRC). Agricultural monitoring is also embedded in the European Union (EU)'s Global monitoring for environment and security (GMES) initiative as one of the major policy areas to be addressed. Moreover, agricultural monitoring is also one of the GMES land monitoring core services as implemented within the Seventh Framework Programme GEOLAND2 project.
The MCYFS has traditionally focused on Europe, western Russia, the Maghreb and Turkey for monitoring agricultural production, as well as selected regions in Africa for food security assessment. Recently, the activities of the MCYFS have broadened and now include the general monitoring of crops in all of Russia, central Asia, China and some middle-east countries. This broadening of the MCYFS towards Russia, central Asia and China has consequences as these regions are generally characterised by more continental climates with harsh winter conditions and warm and dry summer conditions.
Particularly winter crops are affected by prolonged periods of low temperatures during the winter. A feature of cultivation of winter wheat is that its growth starts in autumn and proceeds again in spring after a dormant period in winter. Thus, conditions during autumn and winter determine whether rapid regrowth is possible in spring and have a considerable impact on the final crop productivity. In the majority of the models used for operational monitoring of growth of winter wheat, the impact of autumn and winter conditions on its vegetative state are poorly considered or are not considered in a biophysical approach. Similarly, the crop simulation model which is used within MCYFS (e.g. WOFOST) was calibrated for European conditions and the model is not tailored for continental climates. Particularly the effect of frost damage to winter crops is currently not properly included in WOFOST.
Remote sensing data provide an opportunity to derive information on key crop growth parameters with high spatial resolution. In this case the remote sensing data can be used for an objective estimation of wheat conditions after the winter dormant period. Although the exact amount of standing biomass will be difficult to retrieve directly, it has a large impact on the shape of the leaf area index profile later in the season which can form a basis for real-time updating of wheat growth parameters in the model. Besides, during the crop growth modelling the spatial extent of snow cover and the definition of an exact start date of crop growth is important. This information also can be successively retrieved now with remote sensing data. Finally, remote sensing data can be used to create crop masks of winter-wheat cultivation, particularly in areas with relatively homogeneous land cover as is the case in many of the Russian oblasts.
The overall objectives of MOCCCASIN are to tailor and improve the current MCYFS approach for application in regions with a continental climate, with a focus on winter wheat in the Tula region of Russia. The major components in the project are the following:
1. the collection of field data on Russian winter-wheat varieties and growth conditions (WP2);
2. the within-season mapping of winter-wheat fields using satellite data (WP3);
3. the review, implementation and evaluation of models for frost impact on winter-wheat (WP4); and
4. the retrieval of crop biophysical variables from satellite time series for assimilation into the WOFOST crop model (WP5).
These improvements should lead to improved quantitative prediction of winter-wheat yield, as well as improved (qualitative) spatial assessment of frost impact.
Project results:
For the field work task, partners IKI and AMI initiated meetings with representatives of selected collective farms in the Tula region near the towns of Odoyev and Plavsk. Agreements were made with the collective farms regarding access to the farm and the destructive samplings that were to be performed during the field visit. During the 2011 winter-wheat growing season, the first visits to the test sites were carried out at the end of May. Over the whole growing season 12 field visits were carried out during which 23 fields were sampled for biomass, canopy leaf area index and crop spectral response.
Work package three (WP3) winter crop mapping is the most mature aspect in MOCCCASIN and considerable progress has been made towards the project objectives. First of all an overview has been made of existing approaches for winter-wheat mapping using medium resolution satellite data. This document demonstrates the operational feasibility of the operating chain developed at IKI. Next, the processing chain was applied to generate masks of winter-wheat for the Tula region for the 2002 / 2003 season up till the 2011 / 2012 season based on the MODIS imagery. Finally, an intercomparison of crop mapping using MODIS, MERIS and KMSS is currently ongoing.
WP4 deals with the retrieval of crop biophysical variables from medium resolution satellite data using radiative transfer models. A literature overview of retrieval methods for crop biophysical variables has demonstrated that the neural network inversion of radiative transfer to obtain crop green area index appears to be the most robust retrieval tool. The retrieval algorithms have been setup for the Tula region and preliminary tests been carried out on a test site near Kireyevsk for which archived local crop maps were available over several years. Currently, the retrieval of the biophysical parameter from MODIS for the whole Tula region is in progress based on the winter-crop maps provided by IKI in the WP3.
WP5 deals with the adaptation of WOFOST to include the modelling of frost impact and to allow assimilation of satellite observations. From literature, the FROSTOL, CERES-Wheat, STICS and ClimCrop models have been reviewed which all contain routines for the assessment of frost damage and winter kill. Based on this review, the approaches implemented in FROSTOL and CERES-Wheat have been recognised as most appropriate for Russian conditions. Review of the requirements of the FROSTOL model indicated that considerable changes were needed to the WOFOST source code. Therefore, it was decided to re-implement WOFOST. Currently, the re-implementation of WOFOST is finished and the FROSTOL and CERES-Wheat winter-kill algorithms have been implemented including additional simulation components that were needed.
Finally, the outputs from the different WPs into a winter-wheat monitoring system were stored. Several important steps have been made towards implementing the system:
1. availability of vector layers with field boundaries of the Tula region;
2. digitising the soil map and connection of the soil type to each field;
3. implementation and processing of the (agro)meteorological data needed for the crop simulation model for the years 2009, 2010 and 2011.
Potential impact:
So far, the MOCCCASIN project has already provided some important preliminary results, such as a database with biophysical properties of Russian wheat cultivars, a time series of winter-crop masks for the Tula region and new components for frost impact modelling. Towards the end of the project, there should be improved methodologies for assessment of frost damage and winter kill in continental climates provided. This should benefit the GMES land cover monitoring core services on agriculture. In practice, it should lead to improved crop yield monitoring in the MCYFS for Europe, although there is interest from the Russian partners to use such techniques for a Russian monitoring system, as well.
In a broader perspective, the MOCCCASIN contributes to the Global Earth observation system of systems (GEOSS) agricultural societal benefit area targeting the components global mapping and monitoring the changes in distribution of cropland area and monitoring the global agricultural production leading to accurate and timely reporting of national agricultural statistics and accurate forecasting of shortfalls in crop production and food supply. Moreover, MOCCCASIN contributes to developing and improving analytical tools and methods for agricultural risk assessment (GEOSS Task AG 07 02) related to winter kill and frost damage.
List of websites: www.mocccasin.eu
Information on the outlook of yield and production of crops over large regions is essential for government services dealing with import and export of food crops, for agencies with a role in food relief and for international organisations with a mandate in monitoring the world food production and trade. In 2007, unbalances in the global production of agricultural commodities due to co-occurrence of shortfalls in harvests caused the marked prices of agricultural commodities to peak. These events sparked widespread concern about global agricultural production. Moreover, increased demand for dairy, meat and biofuel products, which require cereals and oil crops, will increasingly compete with the demand for food crops. Given this background the need for a global monitoring system for agricultural production is undisputed and included in the priority themes for Group on Earth observations (GEO). Moreover, the importance was confirmed by the ministerial declaration of the G20 meeting of agricultural ministers in June 2011.
In Europe, the necessity for monitoring agricultural production and prediction of crop yield has long been recognised and implemented through the MARS crop yield forecasting system (MCYFS) operated by the Joint research centre (JRC). Agricultural monitoring is also embedded in the European Union (EU)'s Global monitoring for environment and security (GMES) initiative as one of the major policy areas to be addressed. Moreover, agricultural monitoring is also one of the GMES land monitoring core services as implemented within the Seventh Framework Programme GEOLAND2 project.
The MCYFS has traditionally focused on Europe, western Russia, the Maghreb and Turkey for monitoring agricultural production, as well as selected regions in Africa for food security assessment. Recently, the activities of the MCYFS have broadened and now include the general monitoring of crops in all of Russia, central Asia, China and some middle-east countries. This broadening of the MCYFS towards Russia, central Asia and China has consequences as these regions are generally characterised by more continental climates with harsh winter conditions and warm and dry summer conditions.
Particularly winter crops are affected by prolonged periods of low temperatures during the winter. A feature of cultivation of winter wheat is that its growth starts in autumn and proceeds again in spring after a dormant period in winter. Thus, conditions during autumn and winter determine whether rapid regrowth is possible in spring and have a considerable impact on the final crop productivity. In the majority of the models used for operational monitoring of growth of winter wheat, the impact of autumn and winter conditions on its vegetative state are poorly considered or are not considered in a biophysical approach. Similarly, the crop simulation model which is used within MCYFS (e.g. WOFOST) was calibrated for European conditions and the model is not tailored for continental climates. Particularly the effect of frost damage to winter crops is currently not properly included in WOFOST.
Remote sensing data provide an opportunity to derive information on key crop growth parameters with high spatial resolution. In this case the remote sensing data can be used for an objective estimation of wheat conditions after the winter dormant period. Although the exact amount of standing biomass will be difficult to retrieve directly, it has a large impact on the shape of the leaf area index profile later in the season which can form a basis for real-time updating of wheat growth parameters in the model. Besides, during the crop growth modelling the spatial extent of snow cover and the definition of an exact start date of crop growth is important. This information also can be successively retrieved now with remote sensing data. Finally, remote sensing data can be used to create crop masks of winter-wheat cultivation, particularly in areas with relatively homogeneous land cover as is the case in many of the Russian oblasts.
The overall objectives of MOCCCASIN are to tailor and improve the current MCYFS approach for application in regions with a continental climate, with a focus on winter wheat in the Tula region of Russia. The major components in the project are the following:
1. the collection of field data on Russian winter-wheat varieties and growth conditions (WP2);
2. the within-season mapping of winter-wheat fields using satellite data (WP3);
3. the review, implementation and evaluation of models for frost impact on winter-wheat (WP4); and
4. the retrieval of crop biophysical variables from satellite time series for assimilation into the WOFOST crop model (WP5).
These improvements should lead to improved quantitative prediction of winter-wheat yield, as well as improved (qualitative) spatial assessment of frost impact.
Project results:
For the field work task, partners IKI and AMI initiated meetings with representatives of selected collective farms in the Tula region near the towns of Odoyev and Plavsk. Agreements were made with the collective farms regarding access to the farm and the destructive samplings that were to be performed during the field visit. During the 2011 winter-wheat growing season, the first visits to the test sites were carried out at the end of May. Over the whole growing season 12 field visits were carried out during which 23 fields were sampled for biomass, canopy leaf area index and crop spectral response.
Work package three (WP3) winter crop mapping is the most mature aspect in MOCCCASIN and considerable progress has been made towards the project objectives. First of all an overview has been made of existing approaches for winter-wheat mapping using medium resolution satellite data. This document demonstrates the operational feasibility of the operating chain developed at IKI. Next, the processing chain was applied to generate masks of winter-wheat for the Tula region for the 2002 / 2003 season up till the 2011 / 2012 season based on the MODIS imagery. Finally, an intercomparison of crop mapping using MODIS, MERIS and KMSS is currently ongoing.
WP4 deals with the retrieval of crop biophysical variables from medium resolution satellite data using radiative transfer models. A literature overview of retrieval methods for crop biophysical variables has demonstrated that the neural network inversion of radiative transfer to obtain crop green area index appears to be the most robust retrieval tool. The retrieval algorithms have been setup for the Tula region and preliminary tests been carried out on a test site near Kireyevsk for which archived local crop maps were available over several years. Currently, the retrieval of the biophysical parameter from MODIS for the whole Tula region is in progress based on the winter-crop maps provided by IKI in the WP3.
WP5 deals with the adaptation of WOFOST to include the modelling of frost impact and to allow assimilation of satellite observations. From literature, the FROSTOL, CERES-Wheat, STICS and ClimCrop models have been reviewed which all contain routines for the assessment of frost damage and winter kill. Based on this review, the approaches implemented in FROSTOL and CERES-Wheat have been recognised as most appropriate for Russian conditions. Review of the requirements of the FROSTOL model indicated that considerable changes were needed to the WOFOST source code. Therefore, it was decided to re-implement WOFOST. Currently, the re-implementation of WOFOST is finished and the FROSTOL and CERES-Wheat winter-kill algorithms have been implemented including additional simulation components that were needed.
Finally, the outputs from the different WPs into a winter-wheat monitoring system were stored. Several important steps have been made towards implementing the system:
1. availability of vector layers with field boundaries of the Tula region;
2. digitising the soil map and connection of the soil type to each field;
3. implementation and processing of the (agro)meteorological data needed for the crop simulation model for the years 2009, 2010 and 2011.
Potential impact:
So far, the MOCCCASIN project has already provided some important preliminary results, such as a database with biophysical properties of Russian wheat cultivars, a time series of winter-crop masks for the Tula region and new components for frost impact modelling. Towards the end of the project, there should be improved methodologies for assessment of frost damage and winter kill in continental climates provided. This should benefit the GMES land cover monitoring core services on agriculture. In practice, it should lead to improved crop yield monitoring in the MCYFS for Europe, although there is interest from the Russian partners to use such techniques for a Russian monitoring system, as well.
In a broader perspective, the MOCCCASIN contributes to the Global Earth observation system of systems (GEOSS) agricultural societal benefit area targeting the components global mapping and monitoring the changes in distribution of cropland area and monitoring the global agricultural production leading to accurate and timely reporting of national agricultural statistics and accurate forecasting of shortfalls in crop production and food supply. Moreover, MOCCCASIN contributes to developing and improving analytical tools and methods for agricultural risk assessment (GEOSS Task AG 07 02) related to winter kill and frost damage.
List of websites: www.mocccasin.eu